From Crowdwork to Professional Graph
Joe Kim, CEO of Office Hours, on the end of crowdwork
The real asset is shifting from a job board or call marketplace into a reusable map of verified expertise that can be sold in many different ways. Office Hours starts with paid expert calls and user research, Mercor starts with AI screening and hourly expert contractors for labs, and Handshake starts with a campus recruiting graph, but all three are building the same underlying infrastructure, who knows what, how well they know it, and how to route demand to that person quickly.
-
The product convergence is concrete. Office Hours is building self serve search, matching, scheduling, compliance, and payments for expert calls, mentorship, surveys, and AI interviews. Mercor uses AI interviews and tests to vet experts, then sells access to them for hourly work or recruiting. Handshake repurposed its university graph into Handshake AI, using PhDs, postdocs, and grads for model training.
-
What changes is the unit of value. Legacy firms like AlphaSights and GLG were built like services businesses, with humans manually finding and booking experts for hedge funds and consultants. Newer platforms treat expertise as data and workflow. Once a platform knows an oncologist, a lawyer, or a physics PhD can reliably answer a certain class of question, that same record can power research, hiring, mentoring, and model evaluation.
-
The economics explain why this matters now. Mercor reached $50M ARR by the end of 2024 and $500M annualized revenue by August 2025. Handshake reached about $280M annualized revenue by August 2025, with $80M from Handshake AI. Those numbers show that the same credentialed talent pool can monetize far beyond recruiting alone, especially as reasoning models require higher skill human input.
The next step is that these platforms stop looking like single use marketplaces and start looking like operating systems for human expertise. The winners will be the ones that verify skill well, reduce friction to near zero, and let one expert profile generate revenue across calls, contracts, hiring, training, and AI evaluation, with each new use case making the graph denser and harder to displace.